CONSTRUCTING WEB ONTOLOGIES INFORMED BY SEMANTIC ANALYSIS METHOD

wafflebazaarInternet και Εφαρμογές Web

21 Οκτ 2013 (πριν από 4 χρόνια και 22 μέρες)

227 εμφανίσεις

CONSTRUCTING WEB ONTOLOGIES INFORMED BY
SEMANTIC ANALYSIS METHOD
Júlio Cesar dos Reis
Department of Information Systems, Institute of Computing, University of Campinas
and Center for Information Technology Renato Archer, Campinas, São Paulo, Brazil
julio.reis@cti.gov.br
Rodrigo Bonacin
Center for Information Technology Renato Archer, Rodovia Dom Pedro I, km 143,6
13069-901, Campinas, São Paulo, Brazil
rodrigo.bonacin@cti.gov.br
Maria Cecilia Calani Baranauskas
Department of Information Systems, Institute of Computing, University of Campinas
Av. Albert Einstein, 1251, 13083-970, Campinas-SP, Brazil
cecilia@ic.unicamp.br
Keywords: Organizational Semiotics, Semantic Analysis Method, Web Ontology Language, Semantic Web.
Abstract: In the context of the Semantic Web (SW) research, recent proposals have explored new approaches for a
more precise representation of the meanings. These proposals attempt to model the information in a more
adequate way, and, at the same time to be compatible with the SW standards. This paper proposes heuristics
for deriving an initial Web ontology (WO) from Ontology Charts (OCs) produced by the Semantic Analysis
Method (SAM).
1 INTRODUCTION
There is a growing need for methods, techniques and
tools to better represent the semantic aspects of the
information available in Web systems. The first
initiatives taken by Berners-Lee et al. (2001) already
aimed at creating a Web that also takes into account
the meanings of information and not just its structure
and protocols. Nevertheless, recent studies point out
that there are still various limitations and problems
regarding technologies coming from the SW
initiative (e.g., Reis et al., 2010).
Modelling approaches for WOs that uses the
(SAM) (Liu, 2000) as a starting point may provide a
more precise representation of the semantics. This
approach proposed here enables to incorporate into
SW ontologies, concerns and possible
representations arising from a Semiotic perspective
(Reis et al., 2010).
Assuming that the Semiotic approach contributes
with improvements in business modelling, it is
plausible to have both: the Organizational Semiotic
(OS) (Liu, 2000) methods with a different and
valuable view of the social context, and a WO
described in Web Ontology Language (OWL) that is
an interoperable SW Standard. As semantic refers to
meanings, and meanings are socially created by
humans, we expect to create a more faithful
computer ontology considering an information
system with a more abstract conceptual model that
can capture the behaviour of the involved agents.
In order to use the outcome of SAM (i.e. the OC)
with languages that describe WOs, it is necessary to
create a procedure that makes possible and explicit
the construction of OWL models from OC. The
objective of this paper is to propose heuristics to
perform this construction. The relations between the
models are mapped, and one model supports the
construction of the other. The paper is organized as
follows: Section 2 presents the Theoretical and
Methodogical Background; Section 3 describes the
heuristics proposed to create WOs aided by SAM;
Section 3 concludes and points out future works.
203

2 THEORETICAL AND
METHODOLOGICAL
BACKGROUND
This section firstly presents an overview of the main
concepts of SAM; then some characteristics and
properties of WOs are presented.
2.1 Semantic Analysis Method
In SAM “The World” is socially constructed by the
actions of agents, on the basis of what is offered by
the physical world itself. The SAM assists users or
problem-owners in eliciting and representing their
meanings in a formal and precise model. In SAM,
the analyst in the role of facilitator specifies the
required system functions in an OC - a graphic
representation of a conceptual model. The OC maps
the vocabulary and the temporal relationships
between the percepts that those words represent and
describes a view of responsible agents in the focal
domain and their pattern of behaviour named
affordances (Liu, 2000).
Affordance, the concept introduced by Gibson
(1977) can be used to express the invariant
repertories of behaviour of an organism made
available by some combined structure of the
organism and its environment. In SAM, the concept
introduced by Gibson was extended by Stamper
(1993) to include invariants of behaviour in the
social world; affordances are social constructs in a
certain social context (Liu, 2000).
Agent is a special kind of affordance, which can
be defined as something that has responsible
behaviour. Agents are affordances that can take
responsibility both for their own actions and the
actions of others. An agent can be an individual
person, a cultural group, a language community, a
society, etc.
Ontological dependency is formed when an
affordance is only possible if certain other
affordances are available. We say that the affordance
“A” is ontological dependent on the affordance “B”
to mean that “A” exists only when “B” does; e.g.:
for a person to be able to stumble, he/she must first
walk; thus there exist an ontological dependency
between to stumble and to walk.
Determiners are properties which are variants of
quality and quantity that differentiate one instance
from another. Determiner are attributes that enable
one to describe an agent or an affordance;
Specialization, agents and affordances can be
placed in generic-specific structures depending on
whether they possess shared or different properties;
Whole-part, an agent or affordance can be part
of other agent or affordance. The part also owns all
the ontological dependencies from the whole;
Role-Name, an agent can have a specific role
depending on the affordance it has.
2.2 Web Ontology
The term ontology in Computer Science is often
used to refer to the semantic understanding (a
conceptual framework of knowledge) shared by
individuals participating in a given knowledge
domain. An ontology is used to formally specify the
concepts and relationships that characterize a certain
body of knowledge (domain). The formal nature of
ontologies makes them amenable to machine-
readability and provides a well-defined semantics
for the defined terms (Kalyanpur et al., 2004).

Figure 1: Ontology Chart for project management (adapted from Liu, 2000:79).
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
204
A WO is usually described by computational
languages based on logic for knowledge
representation and inference. According to the SW
architecture proposed by Berners-Lee et al. (2001),
the ontology description languages are related to
other Web languages such as Extensible Markup
Language (XML), Resource Description Framework
(RDF) and RDF Schema (RDFS). In order to
address interoperability problems and to define a
universal paradigm for web-based exchange of
ontological information, the World Wide Web
Consortium (W3C) created the OWL which became
a W3C Recommendation (W3C, 2004). Using OWL
as a common language, knowledge experts and
application developers can create, modify, link and
import ontologies in a distributed environment.
OWL is an important piece of the future vision of
the Web, the SW.
3 HEURISTICS TO BUILD OWL
ONTOLOGY INFORMED BY
SAM
The first step of our approach is the application of
SAM in the context under study. An OC is created to
be the source model for the transformation to OWL
code. In order to accomplish that, a set of specific
heuristics must be followed to derive an initial OWL
ontology. The proposed heuristics are based on the
basic principles proposed by Liu (2000) and also by
Bonacin et al. (2004). They have proposed a set of
heuristics to construct Unified Model Language
(UML) diagrams from OC; those heuristics were
adapted to our purpose since there are conceptual
and practical differences between UML and OWL.
The OWL differs from UML in their proposals and
some key concepts such as methods and
composition. It is important to mention that by
applying the proposed heuristics does not guarantee
an equivalent ontology in OWL, and even the
transposition of all its properties; instead, it
represents some support to the analyst during the
modeling process. We assume that the analyst
should be in charge of evaluating the results and
determining the priorities (e.g. fidelity, maintenance,
reuse, and so on) over the modelling processes.
The proposed heuristics were classified
following the concepts from SAM such as:
affordances, agents, determiners, role-name, whole-
part, specialization and ontological dependence.
Figure 1 shows an example of an OC used to
didactically exemplify the approach with the
proposed heuristics. The heuristics with examples
from the OC of Figure 1 are presented as follows:
Affordances – During the SAM the world is
mainly modelled by the identification of social
constructions (affordances), while in the OWL the
world is modelled by the identification of classes
and individuals in the world. The presence of an
affordance in the OC suggests a class to be modelled
into the OWL ontology. For instance, considering
Figure 1, by the SAM perspective, a “project” is an
affordance of the society, and by the perspective of
WOs it can be a class with attributes. If the
affordance named “project” was represented in the
OC, this suggests that there is a class in the context,
and probably it is possible to refer to it using the
“project” name. Based on the procedure of
extracting names of classes from nouns, affordances
that are nouns can be mapped to classes in OWL
(i.e. affordances that suggest entities will be classes
in OWL). However, not all affordances are nouns,
for example the affordance “employ” which is a
verb. The affordances named as verbs will be
mapped to object properties in OWL. Thereby the
affordance “employ” will not be a class in OWL, but
it will be an object property named “employ”.
Agents – The heuristic suggest that all the agents
represented in the OC can be mapped as classes in
OWL and as sub-classes of a class called “Agent”.
This is carried out to identify the agents into the
OWL ontology, so all agents from the OC would
inherit the possible characteristics and properties
from the class “Agent”. Thus, agents such as
“person”, “organization”, “department” presented in
Figure 1 will be classes into the OWL ontology.
Determiners – The closest OWL concept to a
determiner is data property, which should be
connected to the appropriated class. As in the
example of Figure 1, the “function” determiner will
be a data property in OWL and its domain will be set
with the class “employee”; while the “dep_budget”
determiner will be mapped to the agent
“department”, then this data property domain will be
the “department” class.
Role-Name – In OC the role-name is always
connected to an agent. Hence from OC to OWL
role-names will be mapped as sub-classes of the
OWL classes that represent the agent in the left side
of the role-name. For instance, considering Figure 1,
the role-name “employee” will be mapped to a class
in OWL, and this class will be a sub-class of the
class named “person”; the same applies to the role-
name “employer”. The relation between the role-
name with the affordance in its right side can be
CONSTRUCTING WEB ONTOLOGIES INFORMED BY SEMANTIC ANALYSIS METHOD
205

better visualized in the transformation rule that may
implements this heuristic.
Whole-Part – There are two situations of whole-
part relation. First: when both the source affordance
and the target affordance are nouns, the affordances
or agents are mapped to classes; then an object
property called “partOf” can be created, and the
target class will be a restriction of this source class.
For example, in Figure 1, the class agent
“department”, mapped as a class in OWL will be
part of the agent “organization”, also mapped as a
class into OWL. In the second situation, both
affordances are verbs, so based on the affordance
heuristic described, both will be object properties in
OWL. Therefore, the target affordance will be
mapped to sub-property of the source affordance,
which is also an object property. Moreover, since
there is not part without the whole, when there is a
whole-part relationship there is also an ontological
dependence between the affordance of the whole and
the part.
Specialization – The specialization can be used
in agents, affordances or role-names; the
specialization relation between the generic and the
more specific type can happen between nouns and
also between verbs, as in whole-part relation. When
the more generic affordance type is an action and it
is mapped to an object property, then the more
specific affordance will be mapped to sub-property
of the object property in the OWL that represents the
more generic affordance. Nevertheless, when the
more generic affordance type is an entity (i.e. an
OWL class), and consequently is mapped to a class
in OWL, the more specific affordances will be
mapped to classes in OWL and they will be sub-
classes of the more generic class. The situation when
the more specific affordances are verbs is an
exception in the OC.
Ontological Dependence – This relation
between affordances is the most common in the OC
modeling. When an object cannot exist without
other, an association between classes can be
modelled into OWL. For example, the ontological
dependence that exists between the affordances
“society” and “person” in Figure 1 suggests an
association between then in OWL. For that, creating
an object property named “depends_on”, the source
affordance can be mapped to the range of this
property, and the target affordance is mapped to
domain of this property. Considering Figure 1, the
affordance “project” is ontologically dependent on
the affordance “organization”; thus the
transformation will create an object property stating
that “project” depends on “organization”. There is a
temporal relation between the ontological
dependence of two affordances; so an affordance
that depends on other will just exist while the other
exists. Nevertheless, just using the object property as
proposed in this heuristic is not enough to fully
represent the concept of ontological dependence of
SAM into OWL. Rules described in Semantic Web
Rule Language may be used to represent it.
4 CONCLUSIONS
The SW evolution depends on methods and
solutions that can adequately represent the
knowledge presented in Web applications content.
Heuristics to support the creation of a WO described
in OWL from the outcomes of the SAM were
presented with this proposal. The solution brings
opportunities to improve the semantic models used
in the existing SW applications. Next steps are the
construction of tools to implement the proposed
heuristics as transformation rules and the conduction
of practical experiments illustrating the application
of this approach.
REFERENCES
Berners-Lee, T., Hendler, J., Lassila, O., 2001. The
Semantic Web, Scientific American.
Bonacin, R., Baranauskas, M. C. C., Liu, K., 2004. From
Ontology Charts to Class Diagrams: semantic analysis
aiding systems design. In Proceedings of the 6th
International Conference on Enterprise Information
Systems, ICEIS 2004, Porto, Portugal. v. 1.
Gibson, J. J., 1977. The Theory of Affordances. In
Perceiving, Acting, and Knowing. Eds. Robert Shaw
and John Bransford.
Kalyanpur, A.; Golbeck, J.; Banerjee, J.; Hendler, J., 2004.
Owl: Capturing semantic information using a
standardized web ontology language, Multilingual
Computing & Technology Magazine, Vol. 15, issue 7.
Liu, K., 2000. Semiotics in information systems
engineering. Cambridge University Press.
Reis, J. C., Bonacin, R. And Baranauskas, M. C. C., 2010.
A Semiotic-based Approach to the design of Web
Ontologies. In Proceedings of the 12th International
Conference on Informatics and Semiotics in
Organisations – ICISO 2010. Reading, UK. pp. 60-67.
Stamper, R. K., 1993. Social Norms in requirements
analysis - an outline of MEASUR. In Jirotka M,
Goguen J, Bickerton M. (eds) Requirements
Engineering, Technical and Social Aspects. New
York.
World Wide Web Consortium (W3C), 2004. OWL-Web
Ontology Language, Recommendation 10 February
2004, <http://www.w3.org/TR/owl-features>,
Accessed January 2011.
ICEIS 2011 - 13th International Conference on Enterprise Information Systems
206